Bump engtagger from 0.4.3 to 0.4.5

[dependabot]

Bumps engtagger from 0.4.3 to 0.4.5.

Changelog

Sourced from engtagger's changelog.

0.4.5 (2026-08-14)

License metadata correction; no code changes. This library is a port (2009) of Perl Lingua::EN::Tagger, which has been licensed GPL-3.0-only since its version 0.14 (2008). As a derivative work EngTagger is therefore GPL-3.0-only: the bundled LICENSE is now the GPLv3 text and the gemspec declares GPL-3.0-only. Earlier releases carried a GPLv2 license text, and 0.4.4 declared GPL-2.0-or-later; both were inaccurate.

0.4.4 (2026-08-14)

This is the final planned release of the 0.4 line. Version 2.0 replaces the HMM engine with an averaged perceptron and does not ship the Penn Treebank-derived probability tables, so its tagging output differs. Pin gem "engtagger", "~> 0.4" to stay on this line; it will receive critical fixes only.

• Fix the test suite silently rewriting the bundled lexicon files (lib/engtagger/*.hash) on every run: a path-join bug in the test setup made the existence check always fail, so install ran before each test. This was also the enabler of the 0600-permission contamination behind #20. Tests are ~6x faster as a result.
• get_sentences now applies its intended punctuation cleanup. Four of its five reattachment rules had discarded their results since the original port (only the leading-backquote rule was ever applied), and their regexes were corrupted by string escaping (\W matched a literal "W"). Sentences returned by this method now read naturally ("Lisa Raines, a lawyer,","$3"). Known limitation carried over from the tokenizer: contractions other than possessives remain space-separated ("I 'm", "is n't").
• Fix the public WORD constant, whose regex was corrupted by the same escaping problem and matched only <w>-style tags.
• Tag regexes produced by EngTagger.get_ext (NUM, NN, SEN, etc.) now end with a real \s*, so they consume trailing whitespace including newlines instead of spaces only.
• assign_tag no longer raises NoMethodError for words absent from the lexicon; it returns "" so callers fall back to "nn" as designed.
• EngTagger::VERSION is now defined at runtime (engtagger/version was not required by the library itself).
• Read and write the Marshal lexicon files in binary mode to avoid corruption on Windows.
• Remove the stray shebang and executable bit from lib/engtagger.rb, and the obsolete require "rubygems".
• Clarify licensing: the LICENSE file had a stray MIT license appended after the GPLv2 text. The library is licensed as GPL-2.0-or-later; the MIT tail has been removed and the gemspec now declares the SPDX identifier.
• Gemspec: use the https homepage URL, add metadata URIs, and drop the obsolete test_files.
• Fix a dead test (text_get_ext typo) that had never run and would have failed if enabled.
• Note in the README that 0.4.x carries Penn Treebank-derived probability

... (truncated)

Commits

• 047cbf5 Correct license metadata to GPL-3.0-only for 0.4.5
• 637faa2 Prepare 0.4.4: fix long-standing latent bugs, clarify licensing
• 7630fee Ignore local experiments directory
• See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)